From Parameters to Performance: A Data-Driven Study on LLM Structure and Development
This work provides data-driven insights to guide targeted development and application of future LLMs, addressing a gap in the field but is incremental as it builds on existing knowledge without introducing new methods.
The authors tackled the lack of systematic research on how structural configurations affect LLM performance by creating a large-scale dataset of open-source LLM structures and benchmarks, conducting data mining to quantify these relationships and validating findings with mechanistic interpretability.
Large language models (LLMs) have achieved remarkable success across various domains, driving significant technological advancements and innovations. Despite the rapid growth in model scale and capability, systematic, data-driven research on how structural configurations affect performance remains scarce. To address this gap, we present a large-scale dataset encompassing diverse open-source LLM structures and their performance across multiple benchmarks. Leveraging this dataset, we conduct a systematic, data mining-driven analysis to validate and quantify the relationship between structural configurations and performance. Our study begins with a review of the historical development of LLMs and an exploration of potential future trends. We then analyze how various structural choices impact performance across benchmarks and further corroborate our findings using mechanistic interpretability techniques. By providing data-driven insights into LLM optimization, our work aims to guide the targeted development and application of future models. We will release our dataset at https://huggingface.co/datasets/DX0369/LLM-Structure-Performance-Dataset